General Updates | AAV vector integration events and malignancy surveillance in hemophilia gene therapy recipients

Recombinant AAV vectors are usually described as non-integrating vectors because they form episomes and they don’t integrate to the same extent as viral vectors known for integration, such as retroviral vectors. But nevertheless, AAV vectors also integrate at a frequency estimated between 1 to 32 integration sites per 2,000 cells -we have seen that in animal studies and in clinical trials. And this may seem low, but there are about 100 billion hepatocytes in the liver, so even those low rates of integration mean that there are likely millions of integration events in people who receive clinical doses of vectors. And now there is no evidence they can trigger oncogenesis in humans. The only evidence we have comes from mice, which are genetically prone to develop cancer in response to AAV. But we have not seen malignancies triggered by AAV in other animals, and we have not seen evidence for AAV causality in clinical vector recipients who developed cancer. 

So why do we care about those integrations then? Well, because we know that genome integration of retroviral vectors can trigger oncogenesis, although the mechanisms of those integrations are different. So we know recombinant AAV can integrate, and that it might have implications, and that there is a theoretical risk of tumorigenesis. And if we acknowledge the numbers involved that I mentioned earlier and observations across the board, they warrant vigilance. And that means we need to take a closer look whenever AAV gene therapy recipients develop cancer. 

In clinical gene therapy using AAV vectors, there have been a total of eight cases of cancer. And importantly, none of those cases produced evidence that AAV integration was involved in tumorigenesis. Also, only one of those was hepatocellular carcinoma. This contrasts with mice, which predominantly develop HCC. But it’s important to remember that despite the liver tropism, AAV vectors delivered intravenously go everywhere, and although they might not be able to induce transgene expression in those off-target tissues due to tissue-specific promoters, they may still enter cells in those off-target tissues and integrate in the genome. And that’s why malignancy surveillance should and does extend beyond the liver. 

Of the eight cancer cases, six occurred in people with hemophilia and two in children with spinal muscular atrophy. Importantly, they use different approaches to analyze vector integrations, and all these methods have their strengths and weaknesses. So even though all those studies concluded that there was likely no relationship between tumorigenesis and vector integration, they did not come to those conclusions with the same level of confidence, if you will. Vector integration analysis methods have their sample quality requirements, so tumor sample quality is really what determines the scope of analysis and how meaningful and conclusive it is in terms of causality assessment. No method is perfect, and ideally a molecular analysis should involve several methods that provide complementary information and may provide a broader understanding of individual cancer cases. 

Good communication between the reporting physician, the pathologist, and the manufacturer is key on that front. Approved products list contact information for reporting malignancies, and they provide instructions on collecting samples for integration analysis. Of course, cancer treatment will have priority over everything else, but if sample collection is feasible without compromising patient management, it should be pursued. I also think that the cancer cases reported so far show the need to better define communication goals when cancer happens. The conclusion that cancer was unlikely to be related to gene therapy meant something else in those individual cases. It can mean that no integrations were found. It can mean we found integrations, but they did not seem to have triggered tumorigenesis. Or it can mean that we did not find anything, but our sample quality was insufficient for comprehensive integration analysis. And it can mean several other scenarios, really. So I think that for the benefit of the community, we need to find a way to translate the complex outcomes of molecular analyses to something accessible and informative enough at the same time. So we land somewhere in the middle between too complex for the broader community and too simple to be helpful. And we need a long-term follow-up since tumorigenesis may occur years or decades after a vector infusion. So we need a global approach to data collection since robust analysis of rare events in people with rare diseases will require meaningful numbers of patients. So data should be collected in a global registry, such as the World Federation of Hemophilia Gene Therapy Registry.

This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.